N86-17841
OVERVIEW OF SERI'S HIGH EFFICIENCY SOLAR CELL RESEARCH
John P. Benner, Lee A. Cole and Cecile M. Leboeuf
Solar Energy Research Institute
Golden, Colorado
The general level-of-interest in high efficiency terrestrial solar cells is increasing. Projected
efficiencies of more than 20% are now considered attainable, not only in GaAs based cells, but
also in multijunction amorphous and polycrystalline devices. As III-V solar cells approach this
high performance level, increasing concern is directed toward questions regarding large area
production potential. SERI's program will increase research emphasis on the study of
mechanisms involved in growth of III-V semiconductors in order to develop answers to these
questions.
INTRODUCTION
In 1983, the U.S. Department of Energy established the Five Year Research Plan for the
National Photovoltaics Program. (1) The objective of this plan is to perform the high risk
research needed to establish a technology base from which industry can develop photovoltaic
systems for central station applications. The targeted performance of such installations is to
provide power to the grid at a cost of less than fifteen cents per kilowatt hour, (thirty-year
levelized cost). One approach to achieve this goal is to greatly increase the efficiency of flat
plate and concentrator solar cells. The plan contains milestones to achieve efficiencies of
20% in thin-film gallium arsenide (GaAs) solar cells in 1986 and to reach 35% efficiencies in
multijunction concentrator cells in 1988. In order to achieve these goals, research is needed to
improve the quality of the III-V semiconductor crystal layers and to improve the solar cell
structures to compensate for less-than-ideal semiconductor properties. The High Efficiency
Concepts Task at the Solar Energy Research Institute (SERI) supports research to achieve
these milestones.
Currently, the task supported research can be grouped into three different techniques for
preparation of semiconductor layers. These are growth on low cost substrates which typically
results in polycrystalline layers; growth of single crystal thin films and separation from the
substrate; and heteroepitaxial growth of GaAs or ternary alloys on GaAs or silicon. Each of
these approaches presents some difficult problems for the crystal grower and device designer.
Polycrystalline Gallium Arsenide
The first approach has so far proved to be most difficult. The films are generally
polycrystalline with an average grain size less than a few millimeters. Achievement of this
grain size requires various recrystallization processes. This can result in segregation of
impurities and possibly precipitates at the grain boundaries.(2) The grain boundaries may then
provide shunt paths reducing the performance of the cells. Several studies of films prepared
with impurity-free grain boundaries have shown the validity of the double depletion layer
model for polycrystalline GaAs.(3,^) This model suggests that even clean boundaries will be
detrimental to solar cell performance. One study has shown that the intragrain properties of a
few defective grains may be the dominant cause of poor performance in some solar cells.(4)
Thus, the approach using low-cost substrates presents several challenging problems. Some
23
/
PRECEDING page blank NOT FILMED
8 f f - 3 <: W
topics of interest include grain size enhancement or formation of single crystal films on low-
cost substrates; passivation or neutralizaton of grain boundaries; doping of polycrystalline
films; and development of device structures which minimize the detrimental effects of the
non-ideal films.
Single Crystal Thin Films
The second approach, relying on reuse of a more expensive substrate which promotes single
crystal growth, has shown more success. Using lateral overgrowth of a masked GaAs
substrate, with film separation by controlled cleavage, thin film cells have achieved reported
efficiencies of nearly 19%.(5) Continued research on this approach (termed Cleavage of
Lateral Epitaxial Films for Transfer or CLEFT) is expected to achieve the 1986 DOE Five
Year Plan milestone for thin-film gallium arsenide. Other techniques for separation of single
crystal films heteroepitaxially grown on low melting point or selectively etchable layers
provide promise of useful alternative technologies.(6,7) Another opportunity for this general
approach is the separation of high efficiency cells having a bandgap of approximately 1.75 eV
which can be mechanically stacked or silicon solar cells to form a very high efficiency
optically cascaded stacked concentrator cell or flat plate module. Given that the films are
single crystal, the device design and development are somewhat more straight forward.
However, control of the solar cell's thickness may yield higher performance than is obtainable
in a bulk device. (8) This area will benefit from research on alternatives to the CLEFT process
for growth, separation and handling of thin single crystal films.
Heteroepitaxy
The final area, heteroepitaxial growth of single crystal layers, covers both growth of ternary
and quaternary alloys on GaAs substrates for concentrator cells as well as growth of III-V
semiconductors on silicon or germanium-silicon substrates. The preparation of monolithic
multijunction cells of very high efficiency is attractive for both concentrator and flat plate
modules due to the simplicity of interconnection in the overall system. They also offer
potential for lower optical losses and fewer problems with removal of heat. With the
exception of the AlGaAs/GaAs system, heteroepitaxial III-V systems introduce problems of
control of lattice misfit dislocations and, in some systems, mismatch of thermal expansion
coefficients. Various techniques for composition grading and superlattices are under study to
provide control of propagation of dislocations.(9, 10) The use of controlled strain between
layers is seen to minimize propagation of dislocations. However, in the case of growth of
GaAs solar cells on silicon substrates, it is suggested that the strain induced by dopants
forming the p-n junction actually causes dislocations to bend over at the junction. (11) This
would place the highest density of recombination centers in the space charge region. This
reasoning would explain the lower measured open circuit voltage and fill factor than would be
expected from the observed defect density at the surface of the sample. In systems with
mismatched thermal expansion coefficients, some samples will develop micro-cracks upon
cooling from growth temperatures. These problems can be best addressed by joint efforts in
crystal growth and device design.
Basic Studies
In addition to the specific problem areas listed above, the High Efficiency Concepts Task at
SERI is interested in several other general research problems. Most of the current efforts use
metalorganic chemical vapor deposition (MO-CVD). Several studies have identified impurities
24
in current source materials which create electrically active defects (12,13) in the resulting
semiconductors. MO-CVD is still a relatively new crystal growth technology with new
generations of potential source materials still being introduced. There is considerable room
for improvement in the understanding of the chemistry of the reactions which result in growth
of the crystals and production of effluents. Analysis of gas flows and of source depletion is
important for the development of an analytic approach to design of CVD reactors.
Reducing the temperature required to grow high quality III - V semiconductors may also
provide an important tool for achieving high efficiency solar cells. A wider range of allowable
growth temperatures may provide greater control over the strain in heteroepitaxial crystals.
This may be important for minimizing dislocation density and microcracks in III-V layers. The
predicted existence of superalloys has also generated increased interest in low temperature
crystal growth. Alloys of III - V binary compounds are generally thought to exist only as
random metastable systems. Research at SERI has shown that minimization of the total
quantum mechanical energy of ordered phases of alloys predicts that, if grown at low enough
temperatures (but with sufficient surface mobility), stable ordered intermediate phases
superalloys" would form, e.g. ordered phases of GaInP 2 , Ga^AsP*, etc. Relative to random
alloys of the same composition these superalloys would have the same lattice constant,
somewhat larger bandgaps, and significantly higher carrier mobilities, and would be
thermodynamically stable. These new materials may be very valuable for reaching new levels
of photovoltaic efficiency.
As the high efficiency cell technology begins to approach the limits imposed by the best
materials, increased attention will be placed on further developing "tools" for device designers
to optimize cell performance. Measurements of critical electronic parameters of III-V
semiconductors present new areas for research. Studies of potential techniques for passivating
surfaces (and grain boundaries) could improve efficiencies. Studies of techniques for
interconnecting top and bottom cells of a monolithic tandem device will also be needed.
Research to improve light collection and improve open circuit voltage and fill factor to bulk
recombination limits will be essential for achieving and exceeding the efficiency goals of the
Five Year Plan.
Conclusions
The bulk of the research efforts supported by SERI’s High Efficiency Concepts area have been
directed towards establishing the feasibility of achieving very high efficiencies, 30% for
concentrator and more than 20% for thin film flat plate, in solar cell designs which could
possibly be produced competitively. The research has accomplished a great deal during the
part two years. Even though the desired performance levels have not yet been demonstrated,
based on the recent progress, a greater portion of the terrestrial photovoltaics community
believes that these efficiencies are attainable.
The program can now allocate a larger portion of resources to low cost, large area deposition
technology. The program is currently shifting greater emphasis on to the study of crystal
growth in order to provide the understanding and tools needed to design a large area process.
25
REFERENCES
1. National Photovoltaics Program: Five Year Research Plan, DOE/CE-0072, available
from NTIS.*
2. S. S. Chu et al, "Large Grain Gallium Arsenide Thin Films", Conf. Record 17th IEEE
PV Specialist, Conference, p. 896, Kissimmee (1984).
3. 3. P. Salerno et al, "Electronic Properties of Grain Boundaries in GaAssA Study of
Oriented Bicrystal, Prepared by Epitaxial Lateral Overgrowth", Lincoln Laboratory
Technical Report 669, May 1984, available from NTIS.
4. M. G. Salerno, W. G. Schaft and D. K. Wagner, in Grain Boundaries in
Semiconductors, eds. H. V. Leamy, G. E. Pike and C. H. Seager (North-Holland, New
York, 1982), p. 125.
5. 3. C. C. Fan, R. W. McClelland and B. D. King, "GaHs Cleft Solar Cells for Space
Applications", Conf. Record 17th IEEE PV Spec. Conf., p. 31, Kissimmee (1984).
6. M. S. Cook, "Method of Peeling Epilayers" U. S. Patent 4, 396, 456, August 2, 1983.
7. A. 3. Shukus and M. E. Cowher, "Fabrication of Monocrystalline GaAs Solar Cells
Utilizing NaCl Sacrificial Substrates", Annual Report SERI Subcontract XE-2-02142-
01, available from NTIS
8. R. P. Gale, 3. C. C. Fan, G. W. Turney, and R. L. Chapman, "A New High-Efficiency
GaAs Solar Cell Structure Using a Heterostructure Back-Surface Field", Conf.
Record 17th PV Spec. conf. p. 1422 Kissimmee (1984).
9. L. R. Lewis, "Advanced High-Efficiency Concentrator Cells", Conf. Record 16th PV
Spec. Conf., p. 584, San Diego (1983).
10. M. W. Wanlass, and A. E. Blakeslee, "Superlattice Cascade Solar Cell" Conf. Record
16th PV spec. Conf. p. 584, San Diego (1982).
11. R-Y. Tsaur et al, "GaAs/Ge/Si Solar Cells", Conf. Record 17th IEEE PV Spec. Conf.,
p. 440, Kissimmee (1984).
12. L. M. Fraas, 3. A. Cape, P.S. McLeod, and L. D. Partain, "Measurement and
Reduction of Water Vapor Contnet in AsHj and PH^ Source Gases Used in Epitaxy
to be published in Vac. Sci. Tech.
13. C. Lewis, W. Dietze, and M. V. Ludowise, 3. Elect. Mat. v. 12, p. 507 (1983).
14. G. P. Srivastava, 3. L. Martins and A. Zunger, to be published in Phys. Rev. B, Rapid
Commun.
* National Technical Information Service, U.S. Department of Commerce, Springfield,
Virginia 22161.
26